I took the redeye back last night from SJC and I assumed by the time I started writing this post I would have distilled ONS 2012 down to a few salient thoughts.  So far, I got nothing.  I have some thoughts, but I assumed I would have developed a set of profound conclusions.  This is the disappointing aspect of ONS 2012 for me; a lot of talent on the stage over two days and not a lot of thought provoking dialog.  I am sure others will see it differently.

Take the Google presentations as an example.  By the press coverage you would like that Google unveiled the warp drive.  Google has been talking about this in bits and bytes for a few years.  I heard Bikash Koley present the machine backbone and diurnal backbone at the LigthReading conference in NYC in 2009 or 2010.  What would be more interesting to me is not what Google has done in the WAN, but what Google is doing inside the data center today!  What are they doing today, not what they did over the past few years.  What Google is working on today is a potential leading indicator, but I also understand that the Google team is more inclined to talk about projects that are completing.  My point is that a real value to the networking community at large would be how Google plans to apply or could apply their OpenFlow WAN in the DC.  Forward thinking potential uses cases, pros and cons, strategies to be hashed out…that would be helpful and intriguing.

I thought there was a huge opportunity for vendors to present thought provoking, disruptive solutions – instead most the presentations were pretty high level with the exception of academic presentations.  Maybe ONS needs a format change from panels to speakers where vendors are given 25-45 minutes to say something meaningful.  ONS could take submissions for subject talks like TED (note I have never been to TED).  It was not all bad, I thought Urs Holzle’s keynote was really enjoyable and John Vrionis had the best presentation of the event because his presentation forced the listener to agree or disagree.  Both of those presentations were bookends of the conference.  Maybe I expect too much, but I would really like to hear some technology and solution debates with the gloves off.  This is not T-ball.

In all I thought the event was an excellent networking event, I just expected to hear more discussion of this versus that and why this is better than that and the reasoning to go along with the argument.

/wrk

* It is all about the network stupid, because it is all about compute. *

** Comments are always welcome in the comments section or in private. ** 

I spent yesterday at the Open Networking Summit (ONS) in CA and will be there today.  Rather than a long missive, here are a few quick thoughts:

1.  GOOG: A very enjoyable and thoughtful presentation by Urs.  What it tells me is with a lot of resources, talented people and a design goal specific to their needs – rather than the needs of many; it is possible to build your own networking device.

2. Nicira: After the Monday article in Wired that surmised a world without Cisco due in part to Nicira, I was disappointed by the presentation.  I did not need a Nicira product pitch, but I did not need to have my time wasted either.  I think that when expectations are raised to a hubris level, there is some responsibility to answer those expectations.

3. CSCO: A few thoughts on Cisco.  I saw a number of people from the investment management business (sell side and buy side) and each one told me they were there seeking insight as to the fate of Cisco.  Cisco also revealed some details about their Insieme investment.  If you read the above article on Nicira in Wired and then read the article on Insieme, I do not think it is too hard a stretch to believe that Cisco would be building an SDN platform that implements OpenFlow in ASICs and they plan to create a world in which OpenFlow controllers and abstraction layers are relegated to 1% of the market.  Cisco products use ASICs for their implementation of OpenFlow and the rest of the world is left to follow the merchant silicon track for OpenFlow implementation.   Just a random thought.

More thoughts after today when I can find the time to write.

/wrk

* It is all about the network stupid, because it is all about compute. *

** Comments are always welcome in the comments section or in private. **